What the work involves
Most days you are reading model output in Korean and deciding whether it is actually correct — not just grammatical, but idiomatic, appropriately formal, and culturally plausible for the speaker and situation described. You'll write reference answers when a model's response falls short, annotate specific error spans (honorific misuse, unnatural calques, register drift, mistranslated particles), and justify your rating in English so reviewers and researchers can follow your reasoning. Other tasks include translating and localizing prompts and responses between Korean and English with meaning prioritized over literal fidelity, reviewing spoken or transcribed material for natural usage, and helping tighten the annotation guidelines when the rubric doesn't cover a case you keep encountering.
What the platform screens for
micro1's screening is AI-led and conversational. Expect it to probe:
- Native or near-native Korean plus English strong enough to write clear, specific rationales
- Concrete evidence of professional language work — translation, localization, editing, teaching at an advanced level, linguistics coursework or research
- Whether you can articulate why a phrasing is wrong (honorific system, speech levels, Sino-Korean vs. native vocabulary choice, spacing and orthography norms) rather than just flagging it
- Rubric discipline: consistency across similar items, and restraint about scoring on personal stylistic preference
- Follow-up depth — the interviewer will push on an example you raise, so bring real ones
No prior AI experience is required, and the screen does not test machine learning knowledge.
Logistics
Fully remote, contractor engagement, largely asynchronous. Volume varies by customer project; contributors commonly commit somewhere between 10 and 30 hours per week, with occasional written or verbal check-ins about progress and edge cases. Pay is stated as observed on the platform in the $65–98/hr band — actual offers depend on credentials, task type, and project, and are not guaranteed. Reliable equipment, a quiet environment for any audio review, and responsiveness to guideline updates matter more than a fixed daily schedule.
